Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:127235
Feedback-ID: i8414410d:Fastmail
Precedence: bulk
MIME-Version: 1.0
Date: Mon, 28 Apr 2025 17:36:32 -0500
To: "php internals" <internals@lists.php.net>
Message-ID: <22c056ae-4f54-408b-9146-748ec0e52878@app.fastmail.com>
In-Reply-To: <39af290b-358a-45f4-9d2d-c76dd4e08125@bastelstu.be>
References: <ffadb6e6-5d16-4b45-a020-4b91565e2ee6@bastelstu.be>
 <CAGBsUrc_wjofCOEuz7MNhuVtpmJJwsatDPCjfAdHH6PQriqgZw@mail.gmail.com>
 <c2369d3c-515e-4b8f-a656-bd1265e9d59e@bastelstu.be>
 <CAGBsUrchWDhYX05ztPLGJx0KzyjspeKhu=K=18OVcZLGvTKB0A@mail.gmail.com>
 <8c35314a-e307-4018-8685-39d7a7923cbe@bastelstu.be>
 <4de85c52-9c04-4e66-9654-af1a377015e1@app.fastmail.com>
 <a918f1d5-7c37-4851-bf11-8dc112f40e05@bastelstu.be>
 <bf44d346-0429-4f5b-9b10-dcd9b8aaca00@app.fastmail.com>
 <39af290b-358a-45f4-9d2d-c76dd4e08125@bastelstu.be>
Subject: Re: [PHP-DEV] [RFC] Throwable Hierarchy Policy for Extensions
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From: larry@garfieldtech.com ("Larry Garfield")

On Mon, Apr 28, 2025, at 4:27 PM, Tim D=C3=BCsterhus wrote:
> Hi
>
> On 4/28/25 23:09, Larry Garfield wrote:
>>>> * Would allowing an extension-tagging interface instead of a base c=
lass be an option?  It still allows for catching "anything thrown by thi=
s extension", which I presume is the goal.  If not, why?
>>>
>>> See the =E2=80=9CChoice of Base Exception=E2=80=9D section in
>>> https://github.com/php/php-src/pull/9220.
>>=20
>> That seems to be about not having a common interface for both the Err=
or and the Exception, which makes sense.  I'm talking about `interface E=
xampleException {}` and `interface ExampleError {}`, instead of `class E=
xampleException extends Exception {}`, etc.
>
> Besides not following the de facto standard (which is what this propos=
al=20
> is trying to codify), I'm also not sure what benefit an interface woul=
d=20
> have over a base exception for the problem we're trying to solve here?=20
> So I can return the =E2=80=9Cwhy (interface)?=E2=80=9D. It would just =
make it tempting=20
> to extend some SPL exception :-)

It's a common recommendation in userland, as it allows implementers to e=
xtend an existing exception (eg, InvalidArgumentException) of their choi=
ce while still being tagged as coming from a given library.  Though I su=
ppose if the policy doc also says to never do that, that becomes an irre=
levant consideration.

--Larry Garfield